The Book of Psalms: Composition & Reception. Ed. Peter W. Flint &Patrick D. Miller. SVT 99. Brill: Leiden and Boston 2005

SEPTUAGINTAL EXEGESIS AND THE SUPERSCRIPTIONSOF THE GREEK PSALTER

ALBERT PIETERSMA

Some years ago I had occasion to write a review of JoachimSchaper's, Eschatology in the Greek Psalter.1 In that review I tookexception to Schaper's view on a number of, what I perceived to be,fundamental issues in Septuagintal exegesis. In this essay I would like tocontinue that discussion, not so much contra Schaper as in the largercontext of Septuagintal hermeneutics.

In the broadest of terms one tends to find the field divided between"minimalists," on the one hand, and "maximalists," on the other. In hisbook Schaper takes particular aim at the so-called Finnish School ofSeptuagint studies, because of its propensity—so Schaper—for "notseeing the woods for the trees." He takes issue with what he regards asits essentially mechanistic view on the Greek translator's role which (toSchaper) entails that a translator is not "in any way . . . influenced by hisreligious and cultural environment," but instead is a "mere medium."2 Ido not myself think that Schaper's assessment of the Finnish School isaccurate or fair, but for my present purposes it will do as acharacterization of a "minimalist" approach to exegesis in the Septuagint.Schaper's own approach, by comparison, might then be characterized asone that "does not see the trees for the woods." That is to say, the Greektranslator is effectively elevated to the status of an author and his workbecomes the same kind of replacement for the original as, for example,an English translation of a novel by Kazantzakis. So Schaper writes inthe introduction to his book,

We shall attempt to look at the Septuagint Psalms not merely from aphilological point of view, but also from the perspective of thehistory of ideas. Tracing the development of early Jewisheschatology . . . and trying to assign to the Greek Psalter its properplace in this development will

1 Joachim Schaper, Eschatology in the Greek Psalter. (WUNT 2. Reihe 76; Tübingen:Mohr Siebeck, 1995). For my review, see Bibliotheca Orientalis 54 (1997) 185-90.2Schaper, Eschatology, 21.

444

give us a fresh view of the importance and the formative power ofSeptuagint texts in early Judaism.3

For my immediate purposes, Schaper's view will do as a characterizationof a "maximalist" approach to exegeting the Septuagint, which entailstaking the Greek Psalter as a free-standing entity with its own message,or rather a (more or less) systematically revised message from that of itsHebrew parent. Essentially the same view has more recently beenadvocated by Martin Rösel regarding the book of Genesis.4 Both RonaldS. Hendel5 and William P. Brown6 have raised strong objection to Rösel'sview.

As I see it, the lines between so-called maximalist and minimalistapproaches are increasingly being drawn more sharply, even though eachside maintains that it recognizes the legitimacy in the other's position.So, for example, in the Rösel versus Hendel & Brown debate, Röselrecognizes that some of the differences between MT and LXX aretextual rather than interpretational. Similarly both Brown and especiallyHendel are quite prepared to grant that the LXX is our earliestcommentary on the Hebrew Bible. The crucial question is, When doesthe translated text give evidence for one or the other? Rösel is, of course,correct in emphasizing that each book must in principle be approacheddifferently, since each translator may be expected to have had his ownmodus operandi. Although a meticulous investigation of the translationalcharacter of each book or translation unit may then give somewhatdifferent results for different units, that scarcely means that we shouldnot try to develop a comprehensive explanatory framework within whichvariation can be accounted for and linguistic oddities (as well as beauty)can be accommodated, both among books and within books. Such aninvestigation, however, must clearly deal with textual variants andtranslational variants at the same time, without confusing them. Hence,methodologically Schaper and Rösel should join forces, so to speak, with"the Finnish School" and Brown and Hendel. In short, the (translated)Septuagint needs to be placed

3 Schaper, Eschatology, 64 M. Rösel, "The Text-Critical Value of Septuagint-Genesis," BIOSCS 31 (1998) 62-70.5 R. S. Hendel, "On the Text-Critical Value of Septuagint Genesis: A Reply to Rösel,"BIOSCS 32 (1999) 31-34.6 W. P. Brown, "Reassessing the Text-Critical Value of the Septuagint-Genesis 1: AResponse to Martin Rösel," BIOSCS 32 (1999) 35-39.

445

within the emerging discipline of Translation Studies, and moreparticularly within Descriptive Translation Studies as a branch of thatdiscipline.

THE CONSTITUTIVE CHARACTER OF THE TRANSLATED TEXT

On the subject of Descriptive Translation Studies (hereafter DTS) Iam heavily endebted to the work of Gideon Toury, one of the leadingscholars in the field.7 Within the DTS branch of Translation Studies,according to Toury, three approaches can be used to address threedistinct but interdependent aspects of any translation. This is ofimportance, since the position a translation is intended to occupy withinthe recipient culture, or sub-culture thereof, has a direct bearing on boththe textual-linguistic make-up of that translation as well as on thestrategies by which a target text is derived from its original Secondly, henotes the process-oriented approach, which focuses on the processthrough which a translation is derived from its parent text, and thirdly,the product-oriented approach which seeks to delineate its textual-linguistic make-up along with the relationships which hold target textand source text together. Great emphasis is placed, by Toury, on theinterdependence of all three aspects, that is to say, function determinesproduct and process but it is equally true that each determines the otherin a bi-directional manner. It may be useful to reproduce here Toury'sgraphic representation (slightly expanded) but to remember that thearrows in it can be made to point in either direction.

The (prospective) systemic position &functionof a translation

(function)determines

↓its appropriate surface realization

(= textual-linguistic make-up(product)

governs↓

the strategies whereby a target text (or parts thereof)is derived from its original, and hence the

relationships which hold them together(process)

7 G. Toury, Descriptive Translation Studies and beyond. (Amsterdam/Philadelphia:Benjamins, 1995)

446

Since the Septuagint (for the most part) is a translation, Toury's studywould seem to be directly applicable to the study of the Septuagint. It isnot my purpose here to deal with this issue in any detail but simply tonote a few of the major implications. If Toury's delineation of descriptivetranslational studies is correct, it follows that the three interdependentaspects he delineates, namely, the position or function of the Septuagintin the Alexandrian Jewish community, the process by which it wasderived from its source text, and the relationships it bears to its Hebrew(and Aramaic) source text, comprise its constitutive character.Differently put one might say that function, product and process areembedded in the text as a verbal-object of the target culture thatproduced it. This thought was already adumbrated, apart from Toury'sdelineation, in Boyd-Taylor's article of 1999, where he wrote,

When a translated text is considered with respect to the historicalenterprise which gave rise to it, its originating Sitz im Leben, itbecomes readily apparent that the verbal character of the documentwill to some extent reflect the socio-linguistic practices proper to thelarger cultural undertaking of which it was a part. We might call thisaspect of the text its constitutive character.8

In a sentence, it can be stated that the constitutive character of theSeptuagint is its interlinearity, i.e. its character as a translated text with apronounced vertical dimension that ties it closely to its original. It istherefore the constitutive character of the text that places constraints onhow that text can be interpreted responsibly. Thus what is beingadvocated here is a theoretically principled approach to the entire textwith clear-cut methodological implications and parameters.

Toury further argues that by definition a translation is target-oriented,that is to say, any and every translation answers a felt need within thehost or target culture or sub-culture, and is cloaked in the language ofthat culture. Thus from this perspective even so-called source-orientedtranslating is fundamentally catered to the target culture and hence atheart target-oriented (e.g. the Greek of the Septuagint remains Greek nomatter how Hebraized it might be perceived to be). He writes,

8 C. Boyd-Taylor, "A Place in the Sun: The Interpretative Significance of LXX-Psalm18:5c," BIOSCS 31(1998) 73. This entire article is an excellent piece of exegesis alongthe lines suggested in this essay. For constitutive character see further NETS xiii-xiv.

447

in an almost tautological way it could be said that, in the finalanalysis, a translation is a fact of whatever sector [of the targetculture] it is found to be a fact of, i.e., that (sub)system which provesto be best equipped to account for it: function, product andunderlying process.9

Again let me bring the Septuagint into the picture. For the Septuagint Itake this to mean that the most secure way of placing it withinHellenistic Greek culture, within Alexandrian Jewish Greek culture (as asub-system thereof), within a certain sector of Alexandrian Jewish Greekculture (e.g. worship, law or education), is through an analysis of the textitself by means of the three interdependent approaches Toury hasdelineated: function, product and process. So, for example, if we findthat the translated text in numerous ways is tied to its original and mightbe said to have a pronounced vertical dimension, which involves a gooddeal of negative transfer from the source text, i.e. violations of thelinguistic code of the target language, that should tell us something aboutits original function. If, on the other hand, we uncover few instances ofnegative transfer, hence few if any violations of the linguistic code of thetarget language but instead perhaps a measure of literary beauty, that tooshould reveal something of its original position within the Jewishcommunity. In other words a text written in vulgar Greek and intranslationese points presumably in a different direction from a textwritten with literary beauty and rhetorical flourish. But more importantlyfor my present purpose, such things have a direct bearing on the questionof interpretation and exposition within it. The constitutive character of atranslated text dictates its own hermeneutics.

SOME NECESSARY DISTINCTIONS

As it happens the title chosen by the editors of the current volumeprovides me with a suitable point to continue. It reads The Book ofPsalms: Composition and Reception. I read that to mean that thecomposition of the psalms (or the book of psalms) and their receptionhistory are, though related, nonetheless distinct issues. Thus thecomposing of a piece of literature is one thing but its history ofinterpretation is quite another. It may be, of course, that what thecomposer deliberately encoded in his composition and what a laterinterpreter decoded from that work turns out to be substantially the samething (as far as we can tell), but that conclusion must needs be quod estdemon-

9 Descriptive Translation Studies, 29

448

strandum. In other words, it cannot be presupposed but must instead bedemonstrated to exist. Hence the burden of proof is on the person thatbelieves that the two are effectively one and the same. In exactly thesame way, the translating of the psalms into Greek is one thing but thereception history of the translated psalms quite another. It is thisdistinction, as I see it, that informs James Barr's argumentation in hisbook The Semantics of Biblical Language10 and which he more explicitlystates in his response to David Hill's criticism of his work. Barr therewrites,

He [Hill] does not make the obvious and necessary distinctionbetween two sets of mental processes, those of the translatorsthemselves, whose decisions about meaning were reached from theHebrew text, and those of later readers, most of whom did not knowthe original . . . .11

Or to cite the general introduction to the recently published NETStranslation of Psalms,

. . . just as the [textual] form of the original text differed [inprinciple] from its later textual descendants, so what the originaltranslator thought his text to mean differed [in principle] from whatlater interpreters thought the text to mean.12

My central interest here is in the original translation, in distinction fromlater interpretations with which the text may have become endowed.Thus the operative thought here is that one and the same text should beassumed to have been understood differently by its originator (author,translator, or redactor) and its subsequent interpreters or exegetes. Thisshould at once be obvious when one reminds oneself that all suchactivity occurs within certain cultural environments and are designed tomeet certain cultural needs.13 The NETS Introduction suggests thedistinction between the Septuagint's constitutive character, on the onehand, and its reception history, on the other.

As I see it, Toury too makes a comparable distinction in DTS whenhe writes,

. . . this principle [namely, that function determines textual-linguisticmake-up] does not lose any of its validity when the position occupiedby a

10 J. Barr, The Semantics of Biblical Language (Oxford: Oxford University Press,1961;repr. London: SCM, 1983).11 J. Barr "Common Sense and Biblical Language," Biblicca 49 (1968) 379.12 A. Pietersma and B. Wright (eds.), A New English Translation of the Septuagint: ThePsalms. (New York and Oxford: Oxford University Press, 2000) x.13 Toury, Descriptive Translation Studies, 12.

449

translation in the target culture, or its ensuing functions, happen todiffer from the ones it was initially 'designed' to have; e.g., when thetranslation of a literary work, intended to serve as a literary text tooand translated in a way which should have suited that purpose, isnevertheless rejected by the target literary system, or relegated to aposition which it was not designed to occupy. In fact, one task ofdescriptive studies in translation may well be to confront the positionwhich is actually assumed by a translation with the one it wasintended to have. . . .14

Or again,

. . . significant is the possibility that translations which retain theirstatus as facts of the target culture may nevertheless change theirposition in it over time. Of course, such changes can have no bearingon either the intended, or even the final position of a translation.15

Applied to the Septuagint, I take this to mean that its original andintended function, embedded in its linguistic make-up and in itsrelationships to its parent text, could have differed quite radically fromthe role subsequently assigned to it. More concretely let us suppose for amoment that the Septuagint did begin its existence as a study-aid for theHebrew (thus a crib), as has been suggested elsewhere,16 certainly asearly as Aristeas its position was that of an independent text, a freestanding entity, holy scripture. Likewise for the writers of the NewTestament it was itself holy writ. But assuming for the sake of argumentthat such a development indeed took place, would one then have toconclude as well that its constitutive character had undergone a changecommensurate with its change in position or status? To me that issue isscarcely subject to debate.

It was the failure to draw a distinction between the constitutivecharacter of the Septuagint, on the one hand, and its reception history, onthe other, that in my judgment vitiated much of Joachim Schaper's bookon the Greek Psalter.

I proceed to make a second distinction which is fundamental to myoverall argument, and that is the distinction between the original text ofthe Greek translation and subsequent and therefore secondary changesintroduced into that Greek text. The point I wish to make is this: if oneintends to focus on the original Greek text, i.e. its constitutive character,in order to determine its exegetical dimension vis-

14 Toury, Descriptive Translation Studies, 14.15 Toury, Descriptive Translation Studies, 30.16 E.g., A New English Translation (NETS Psalms) ix.

450

à-vis the Hebrew parent text, whatever can be shown to be secondary tothe pristine text ceases to be grist for the mill. Differently put, secondarydevelopments in the Septuagint belong ipso facto to its history of textualtransmission and its history of interpretation, i.e. its reception history,and consequently are not part of its constitutive character. Yet again,secondary elements may tell us a great deal about how the Septuaginttext was understood at some point in its long transmission history, withina certain cultural setting; they can tell us nothing about the understandingof the translator himself. As a result, the first thing a modern interpreterof the Septuagint must do is to determine what is primary and what issecondary, whether through private research or through reliance on acritical edition. Needless to say, the labels "primary" and "secondary" arenot indicative of ontological status; rather they simply mark logical andchronological precedence and subsequence. Here then, the operativethought is that a given text may be added to or subtracted from, with theresult that a new text, a (slightly) different entity, may be created in theprocess. My interest lies with the first text.

Finally under the present sub-heading, I must briefly return to Rösel,since in a recent article he has gone even farther than he did in the piece Inoted earlier. Whereas in the earlier article in BIOSCS he alreadysuggested, as Hendel rightly noted, that since the parent text ofSeptuagint Genesis and our present Hebrew text were substantially thesame, where they differ must then be interpretational (rather thantextual), in his article on the superscriptions of the Septuagint Psalter17 heboldly asserts that variants in the Septuagint without external attestationshould only be taken to be textual variants if (a) they cannot be explainedas intra-textual harmonization, (b) as being linguistically motivated or (c)as exegetically motivated. In so doing, says Rösel, "soll der Eigenwertder griechichen Übersetzung gegen oftmals naive Textkritik stärkerpointiert werden." What troubles me about Rösel's assertion is not somuch that it prohibits facile recourse to difference in parent text (thoughit seems overly restrictive), but that it seems to suggest that whatever canbe regarded as exegetical should be so regarded. To be sure, Rösel'spoint (b) ("linguistically motivated deviation") might provide animportant escape hatch; yet I read him to say that all differences betweenthe Hebrew text and the Greek text

17M. Rösel, "Die Psalmüberschriften des Septuaginta-Psalter," in E. Zenger (ed.), DerSeptuaginta-Psalter (Freiburg: Herder, 2001) 125-48, esp. 125.

451

are interpretive in nature—until proven otherwise. If that is indeedRösel's stance, I fear that the cart is being put before the horse, and thatall carts and all horses are of the same colour. Rather than working froma text's constitutive character, beginning with what Toury calls processand product, that is, its textual-linguistic make-up and the relationshipsof the translated text to its source, we are effectively advised to workfrom the outside in. This cannot be justified, it seems to me, unless onemaintain that textual-linguistic make-up and relationship to the source, inother words, the vertical dimension of the translated text, have norelevance for exposition. Furthermore, it can scarcely be maintained thatall interpretation is exposition or exegesis—but more on this below. Infact I would formulate the precisely opposite postulate that would run asfollows: No difference between the Hebrew and the Greek texts shall bedeemed exegetical, until proven so. An excellent set of eleven postulateson Septuagint exegesis has recently been developed (with graphicrepresentation) by Frank Austermann.18 His delineations also serve verywell to place the Greek Psalter in descriptive translation studies, and thusto establish a general framework within which it should be studied.Unfortunately, Austermann's argumentation seems to have beensummarily dismissed by Rösel.19

TRANSLATION AS INTERPRETATION

That translation is, and can only be, interpretation rather than beingsimply a reproduction (of the parent text) I do not consider to becontroversial.20 If that is correct, the issue on which I want to focuscannot be whether interpretation occurred when the Hebrew psalms weretranslated into Greek but what level of interpretation was achieved inany given instance. And that in turn leads to a further question: Is itmeaningful to count each and every level of such interpretation asexegesis or exposition? Even an elementary definition of the term in

18 F. Austermann, "Thesen zur Septuaginta-Exegese am Beispiel der Untersuchung desSeptuaginta-Psalters," in A. Aejmelaeus and U. Quast (eds.), Der Septuaginta-Psalterund seine Tochterübersetzungen. (Göttingen: Vandenhoeck & Ruprecht, 2000) 380-86.For a more discursive discussion see Boyd-Taylor “A Place in the Sun,” 71-77.19 M. Rösel, " Psalmüberschriften," 126-27.20 See, for example, Hans-Georg Gadamer, Truth and Method, (New York: Crossroad,1986) 345-66.

452

Webster's New Twentieth Century Dictionary (1956) suggests otherwise.According to WNTCD exegesis is: "the exposition, critical analysis, orinterpretation of a word, literary passage, etc., especially of the Bible." Itwould thus be fair to say, it seems to me, that exegesis, in anymeaningful sense, presupposes as a minimum (a) deliberateness, (b)methodicalness, and (c) a goodly degree of target orientedness. Unlessall three of these are present it makes little sense, I would submit, even tobegin to speak of exegesis or exposition. In what follows I will thereforeargue that exegesis, since by nature it is contextual, can be said to beginonly at a certain level of interpretation.21 Let me make it perfectly clear,however, that I am not denying that exposition and exegesis exist in theSeptuagint. Instead my interest lies in ways of identifying suchexposition responsibly and scientifically.

Though no translator can realistically choose not to interpret, he candecide whether to make his translation more source-oriented or moretarget-oriented. As Gideon Toury has noted, the seventies of the pastcentury were marked by “extreme source-orientedness” and in his wordsthe “preoccupation was mainly with the source text and with theproclaimed protection of its ‘legitimate rights’.”22 This source-orientedness is then contrasted with target-orientedness, without anysuggestion that the two are mutually exclusive. In fact I have notedearlier that, for Toury, at a deeper level target-orientedness includessource-orientedness, since by definition a translation is aimed at thetarget culture. The terms themselves are very helpful since they tell usmuch about a translator's modus operandi and by extension at what levelof interpretation one should understand him. Sebastian Brock,23 inapplying these concepts directly to Greek biblical translation fromHebrew speaks of the difference between, on the one hand, translationsthat bring the reader to the text and, on the other hand, translations thatbring the text to the reader. No doubt the most extreme example ofsource-orientedness within the biblical corpus is Aquila, but from Aquilaone can draw concentric circles to the rest. Thus the operative thoughthere is that the degree of source- or target-orientedness

21 Similarly, Webster's defines hermeneutics as "the science of interpretation, or offinding the meaning of an author's words and phrases and explaining it to others;exegesis: particularly applied to the interpretation of the Scriptures."22 Descriptive Translation Studies, 24.23 S. Brock, “The Phenomenon of the Septuagint,” OTS 17 (1972) 17.

453

of a translation stands in direct proportion to its level of intelligibility, orlack thereof. Whereas Aquila is difficult if not impossible to understandwithout the help of the Hebrew, for Job it is not infrequently advisable toignore the Hebrew.

LEVELS OF INTERPRETATION

But if it is correct, as I have suggested, that not all interpretation canbe called exegesis or exposition, it will be necessary to differentiate.Accordingly, in what follows I will delineate what I have called "levelsof interpretation" and illustrate each, as much as possible, with examplesfrom the superscriptions of the Greek Psalter. My reason is simply that inan earlier article24 on them I have already raised the question of theirinterpretive function, and because the superscriptions furnish me withreasonably good examples for most of what I want to illustrate. Alllevels or categories are, I believe, applicable to any part of the(translated) Septuagint.25 Furthermore, as I noted earlier, Martin Röselhas just written an article on the superscriptions. Since his approach toSeptuagint exegesis is different from my own, I can productively interactwith what he has written. As will become clear, my basic disagreementwith Rösel does not lie so much in the interpretation of individual wordsand phrases as it does in the contextualizing that he proposes. Sinceexposition and exegesis are by their very nature a matter ofcontextualization, my interest in his article should be obvious. As anaside, I might yet note that I tend to read the superscriptions ratheratomistically as a series of notes added over a long period of time.

Level 0: "Interpretation" by TranscriptionThe numbering here is deliberate since items of language transfer

which I place here are not interpretational in any meaningful sense of theterm, since this category is comprised not of just any transcriptions fromthe source language but of transcriptions that had no prior linguisticstatus in the target language. Thus what I have in mind here are not itemslike a9llhlouia/, which in all probability had a history of

24 A. Pietersma, "Exegesis and Liturgy in the Superscriptions of the Greek Psalter," inB. A. Taylor (ed.). X Congress of the International Organization for Septuagint andCognate Studies, Oslo, 1998. (SBLSC 51; Atlanta: Society of Biblical Literature, 2001)99-138.25 It should also be noted that to begin the levels of interpretation effectively at theword-level appropriately reflects G's segmentation.of his source text.

454

usage in Alexandrian Jewish Greek and—if that is so—had beenintegrated into the living language before the translation process began,but items that were transcribed de novo as products of the translationprocess. In fact, one can place here all indeclinable, transcribed names(or any Hebrew lexemes treated as names), instances of which arefurnished aplenty in the superscriptions of the Psalter. From thesuperscriptions which are undeniably original I include the following:Abessalom (3), Abimelech (33, 51, Aithan (88), Asaph (49, 72-82),Bersabee (50), Chousi (7), Dauid (3 et passim), Doek (51), Haiman (87),Idithoun (38, 61), Iemeni (7), Kore (41, 43-46, 83, 84, 86, 87), Nathan(50), Saoul (17, 51, 53, 56, 58), Soba (59). Since such transcriptions intoGreek had no prior history of usage, they lacked reference in Greek. Asan aside it may be of interest to note that, whereas they typically hadsemantic transparency in the source language, this disappeared in theprocess of translation. Since such items lacked reference in Greek andtherefore cannot meaningfully be called interpretive, I have assignedthem to Level 0.

In translation literature, apart from names, one thinks immediately ofTheodotion who had a penchant for throwing the Hebrew text at hisreader without translating it. Yet the phenomenon is well attested also inthe Septuagint, particularly in Greek Jeremiah. Such transcriptions are,however, in short supply in the Psalter, since its translator insisted onrendering his source text into Greek, whether or not he understood it.Perhaps the best example from the superscriptions is tlhm-l( in 53(52)and twn(l tlhm-l( in 88(87), which is generally taken to refer to atune or chanting pattern to be used with these psalms,26 and renderedaccordingly by the NRSV as "according to Mahalath" and "according toMahalath Leannoth" respectively. The Greek translator in 52 and 87 doesa bit of transcribing and translating. Thus in 52 he comes up with u9pe\rmaeleq and in 87 with u9pe\r maeleq tou= a0pokriqh=nai, deriving twn(from the verb hn( ("answer"). Since in Gen 28:9 Maeleth is a daughter ofIshmael, it is not impossible that the Psalms translator intended areference to that person, although the press the lady gets in Gen 28 is notconducive to being mentioned in Psalms superscripts, nor is such aconnection made by the Church Fathers,27 who instead interpreted maeleqas

26 M. E. Tate, Word Biblical Commentary: Psalms 51-100. (Dallas: Word Books,1990.)27 So e.g. Athanasius Expositiones in Psalmos 27.248, Didymus the Blind, Fragmentain Psalmos 868, Eusebius, Commentaria in Psalmos 23.453, and Gregory of Nyssa, Ininscriptiones Psalmorum 5.74.

455

xoro/j ("dance") or xorei/a ("dancing"), gleaned from Aquila,Symmachus and Theodotion, who in turn derived it from Hebrew lwx("whirl/dance/writhe"). Of course, even if the Fathers had connectedmaeleq with Genesis 28, one would simply note it as a fact of receptionhistory rather than of the original text. In Psalms 52 and 87, as elsewherein the superscriptions, our translator shows no knowledge of cultic orliturgical directives. What later interpreters did with such items might beof interest, but irrelevant to the question posed here. The constitutivecharacter of the text in 52 and 87 is clearly one of pronounced source-orientedness.

Also to be placed at Level 0 are all textual items that can readily beexplained as being due to mechanical error such as misreading of Hebrewletters, haplography, dittography, and parablepsis. If that is correct, noneof these can be regarded as expositional in any way.

Quite clearly, this category of "interpretation" is characterized by thehighest possible degree of source-orientedness, and consequentlydemonstrates most vividly the vertical dimension of the translated text,that is, its highly restrictive relationship to its source.

Level 1: Interpretation at the Word LevelWhat happens here is that a lexeme of the source text (Hebrew) is

replaced by a lexeme of the target text (Greek), though not necessarilyintegrated syntactically and therefore supplied with unmarked inflection(nominative). The difference between Levels 0 and 1 is that whereastranscriptions are without reference in the target language, items at Level1 have an established reference. Differently put, they have meaning butas isolated words cannot be said to convey information. Thus someinterpretation does indeed take place, but clearly at a very elementaryand restricted level. As an eloquent example one may cite, from the so-called Kaige recension, e0gw& ei0mi as a representation of ykn) (the longform of the 1st sing. pronoun) even when it occurs with a finite verb.From the superscriptions one may choose the less obvious yalmo/j as arendering of rwmzm. Since both apparently referred to instrumentalrather than vocal music, it may well be that the difference between themwas minimal, though a close correspondence of this type would have tobe labeled accidental to rather than essential for this category ofinterpretation. Furthermore, if Hebrew rmz can refer to the playing onwind instruments and on string in-

456

struments together with the words, while Greek ya/llw refers solely tostring instruments, one might note that in the transfer from source totarget a restriction of meaning has occurred. That yalmo/j, since it hasan established reference in Greek, can readily be used in syntacticconstituents in explanatory contexts, is of course true, but a separateissue. In the superscriptions it is invariably made to represent rwmzm,and is even used in the phrase e0n yalmoi=j in 4.1, where it is scarcelyintelligible. What the textual-linguistic evidence suggests is a mentalprocess that substituted yalmo/j for rwmzm but not one that deliberatelyrelabeled the piece of literature in question from a rwmzm to a yalmo/j.Moreover, to the extent that it did refer to the psalm as a whole, onewould in any case have to credit the source text rather than the translator.For the moulding of the Greek term to fit its new use, one has to look toreception history. Though in time yalmo/j took on the meaning that"psalm" has in English and other modern languages, there is ampleevidence to show that it did not yet have that meaning in Septuagintaltimes.28 The point here is that in reference to the entire piece in whosesuperscript it appears, it is slightly odd, since the piece in question is adescriptive piece of literature rather than a musical performance onstrings, and its isolate use will become even clearer presently.

Other terms in the superscriptions that fall into the same category arew)|dh/ for ry# $ (44[45], 64[65], 75[76], 95[96], 119[120],121[122]—133[134]) (see further below), u3mnoj for hnygn (6, 53[54],66[67], 75[76]), sthlografi/a for Mtkm (15[16], 55[56], 56[57],57[58], 58[59], 59[60]), ai1nesij29 for hlht (32[33], 144[145], 146[147),proseuxh/ for hlpt (16[17], 101[102], 141[142]) though some of thesecan also be cited under my next category since they are pushed by thetranslator to the phrase level, without explicit warrant in the Hebrew. So,for example, sthlografi/a is preposed with ei0j (except in 15) to formsome kind of purpose (or general reference) phrase, without explicitwarrant

28 See, for example, Amos 5:23. It is further of interest that the Church Fathers stillcontrast yalmo/j and w)|dh as instrumental vs. vocal (e.g. Origen on Psalm 29).29 Rösel glosses this as "Loblied" but this can only be justified if what the Hebrew termis thought to mean ("song of praise") is superimposed on the Greek. Greek ai1nesij asan active verbal noun means nothing more than "praise" or "praising." Contrast, on theother hand, ai]noj in superscriptions Psalms 90, 92, 94.

457

in the Hebrew. Here I simply want to emphasize that interpretation onthe word level does indeed take place but that unless such words arenewly integrated into the context of the translated text at least at thephrase level, they can be said to have meaning but cannot fairly be saidto be expositional. That some of these terms happen to make sense inreference to the entire psalm is a bonus, but not to be confused with whattook place at the constitutive stage. Again, the Greek term is present as areflex of its Hebrew counterpart and not because the translator decidedthat the psalm as a whole could best be so described.

Other items that belong at this level are so-called etymologicalrenderings, i.e. Greek words arrived at not because of contextualconsiderations within the Greek but because an unfamiliar item in thesource text is linked to a familiar item and then translated into the targettext, whether or not it fits the context. Jan Joosten in his article onexegesis in Greek Hosea has placed such items under the descriptiveheading "Giving the Words Their Due".30 An instance from thesuperscriptions is tou= a0pokriqh=nai in Ps 87(88) for twn(l, althoughthis is slightly beyond the present level. The Greek translator (hereafterG) not knowing what it means derives it from hn( ("to answer"). Assuch it would have to be read as a qal construct infinitive with theprefixed preposition l, indicating purpose. Thus, as is G's practice insuch cases, the preposition is glossed by the Greek article in the genitive.Though the Greek infinitive happens to be passive in form, given thenature of the verb, that need not mean that it was intended to be passivein function (so NETS ). Thus G is responsible for two items ofinterpretation: (a) one Hebrew lexeme (twn(l) is rendered into Greekand (b) a second Hebrew (tlhm) is construed as the subject of theinfinitive. Since Ps 87(88) happens to be a prayer (see h9 proseuxh/ mouin v. 3) it is not impossible that v. 3 played a role in the latter process.One strongly suspects, however, that his move was purely on theword/phrase level. That his resultant text created potential for futureinterpretation is doubtlessly true, though the Church Fathers evidentlydid not make use of that potential.

Further, what should be placed on this level of interpretation is what

30 J. Joosten, "Exegesis in the Septuagint Version Hosea," in J. C. de Moor (ed.),Intertextuality in Ugarit and Israel (OTS 40.) (Leiden: Brill, 1998) 62-85, esp. 72-73.Joosten's article as a whole is very useful for its grouping of phenomena and for itsstrictly text-based approach.

458

NETS has labeled (semantic) stereotypes, i.e. Greek words woodenlypaired in the process of translation with Hebrew words often as a closedequation. Since words in different languages seldom, if ever, have thesame semantic range, one-to-one representation can cause problems incertain contexts. Because the superscriptions offer minimal context, nogood example can be gleaned from them. One might, however, cite theCpx–qel- equation in Psalms. Since √Cpx includes the semanticcomponent of "pleasure/delight" but √ qel- does not, the latter does notalways smoothly fit its context (cf. e.g. Ps 1:2).

Finally what should be placed here are translated and partiallytranslated names. From the superscriptions one can cite Zifai=oi forMypyz (53[54]), Mesopotami/a for Myrhn Mr)and Suri/a for Mr)(59[60]), 0Israhli/thj for yxrza) (87[88], 88[89]), Mwush=j for h#$m(89[90]).

Though on this level of interpretation the source text does not playas restrictive a role as on the preceding one, it remains true that itseriously interferes with the target text, even though the target languageis being used. Thus the vertical dimension remains the dominant one.Characteristic at this level of interpretation is that words either have nocontext or stand in tension with their context. An initial way of testingwhether a given item belongs to this category is to determine with whatconsistency it is made to represent its Hebrew counterpart, and to whatextent the context is simply reproduced from the source text.

A fascinating exception, although not in the superscriptions, wouldseem to be dia/yalma, always used as a rendering for Hebrew hls butpossibly coined by the translator of Psalms from the same root asyalmo/j, and functionally adjusted accordingly. Since, like Greekdiau/lion, familiar from drama as a musical interlude on the flute(au0lo/j), dia/yalma evidently indicated a musical interlude on a stringedinstrument, it is never made to stand at the close of a psalm (see Ps 3,23[24], 45[46]).

Level 2: Interpretation at the Phrase LevelAs the minimum unit of information it is perhaps understandable that

at this level the greatest potential for maximalist interpretation comes tothe fore. This is so, no doubt, because a phrase out of context or inminimal context gives inherently ambiguous information. Thus hereagain contextualization is the central issue. In other words,

459

based on the linguistic make-up of the translated text how much newcontextualization can legitimately be attributed to the translator? Is thecontext simply transferred from the source text, or is the context thecreation of the translator, as a result of which the target text can be saidto have a context different from that of the source? Since thesuperscriptions of the Psalter are especially rich in phrases, mydiscussion here will be disproportionately long, though still onlyillustrative.

Level 2.1:From the superscriptions I take two related examples, namely,

yalmo\j w|)dh=j (29, 47, 66, 86, 91) and w)|dh\ yalmou= (65, 82, 87, 107).Needless to say, they occur when the corresponding Hebrew terms,rwmzm and ry#$ stand together. The mental process reflected by thereality of the translated text seems akin to our own, especially whenreading unpointed Hebrew. When two Hebrew nouns stand together onemight infer a bound construction such as "X of Y," especially if there isno context to correct one's mistake.31 Though the words themselves havemeaning, what were the phrases yalmo\j w|)dh=j and w)|dh\ yalmouintended to convey as units of information? Seemingly about as much as"a performance on strings of a song" and "a song of a performance onstrings" would convey in English. Of course, one can massage suchphrases into meaning "an accompaniment of a song on strings" or "asong with accompaniment on strings." But if that is what the translatorwanted to convey, surely he could have done so by using or forming aword such as yalmw|di/a "a singing to a harp."32 Such well-intentionedattempts at making sense of the translator's text, however, miss thenature of the text itself. What happens is that two syntactically unrelatedHebrew words are forged into a phrase, evidently without muchreflection on what the combined pair might mean. Thus for context oneis forced to invoke the vertical relationship of the translated text to theparent text, since the translator's modus operandi wreaks havoc with thehorizontal dimension of the Greek. In other words, the best way toaccount for the Greek text we have is interlinearity, and in this instanceof a rather restrictive variety. Thus, beyond the word level, the onlyinterpretation we have here is that two lexemes of the source text aremade into

31 See further Psalm 75(74) where the two do not stand together but yet are treated as abound construction.32 For yaltw|de/w and yaltw|do/j see e.g. 2 Chron 5.13 and 1 Chron 6.33respectively.

460

a phrase in the target text, irrespective of coherent sense. What contextthere is beyond this level is simply carried over from the source.

Perhaps not surprisingly, though Patristic commentators on thePsalms understand what the two Greek words mean separately andmaintain that allegorically yalmo/j has to do with physical activity whilew) |dh / stands for mental activity, the best they can do for theircombination is to say that it means the two combined.33 To illustrate Icite Origen on Psalm 29:

0Orga/nw| de\ kai\ fwnh=| o9 yalmo\j a0potelei=tai dio\ yalmo\jw)|dh=j e0pige/graptai, dhlwtiko\j w@n tou= dei=n h9ma=j kai\o0rga/nw| dia\ tw~n swmatikw~n kinh/sewn u9mnei=n to\n qeo/n, kai\fwnh=| nohth=|, dia\ tou= to\n nou=n a0nakei=sqai tw~| dhmiourgw~|. |34

(The psalm is performed with instrument and voice. Therefore it istitled yalmo\j w)|dh=j, making it very clear that we must sing hymnsto God with an instrument, through bodily movements, and with amental voice, through devoting our mind to the creator.)

In my discussion of Level 1, I have already called attention to otherphrases that belong to this category, though not all of these are assemantically problematic as those just discussed. Since, however,potential exegesis at the phrase level looms larger in the superscriptionsthan at any other level, it may be useful to discuss the more importantone instances in some detail. Rösel too concentrates his attempts atextensive contextualization at the phrase level.

A similar phrase to those just discussed is ai]noj w)|dh=j in 90, 92, and94 without counterpart in MT. Though it is possible that this phrase wasoriginal in only one of the three psalms and from there spread to theother two, its structure makes it unlikely that it was secondaryeverywhere. Semantically it is even more incongruous than eitheryalmo\j w)|dh=j or w)|dh\ yalmou=. Since both terms refer to vocal musicand both indicate songs of praise, it is difficult to understand what thetwo combined might be intended to convey. If, however, one retrovertsthe Greek phrase into Hebrew on the pattern of either of the otherphrases the text becomes transparent. Thus in all likelihood, ai]noj w)|dh=jtranslates ry#$ hlht, analogous ry#$ rwmzm and rwmzm ry#$.

33 See e.g., Athanasius, Expos. in Psalmos 27, 576; Basil, Hom. super Psalmos 29, 305;Didymus the Blind, Comm. in Psalmos 129; Eusebius, Comm. in Psalmos 23, 680.34 Fragmenta in Psalmos on Ps 74.1.

461

Level 2.2:The use of u9pe\r th=j klhronomou/shj for twlyxnh-l) in Psalm 5.

I begin here with the observable facts: (a) that G did not understand theHebrew word as a musical term; (b) that he derived twlyxnh from √lxn("inherit"); (c) that the y infix made the word into a verbal (hiphil) ratherthan the noun hlxn; (d) that the Hebrew article as well as the tw- endingsuggest a nominalized participle; (e) that as a participle of √lxn it wouldhave to be an active participle; (f) that the feminine inflection of thesource text produced a feminine inflection in the target text; (g) that thestandard gloss in Psalms for √lxn ("inherit") is klhronom- (22x) (1exception). Thus apart from his mistaken identification, which canscarcely count as exposition, the only real expositional move he makes isto construe the Hebrew word as a singular rather than as a plural, a movevery similar to the move we saw him make above in 87(88). Thatinterpretation took place is obvious: a Hebrew word is replaced with aGreek word and, more particularly, an unknown Hebrew word isreplaced with a known Greek word. But given the questionable thoughunderstandable derivation, what G did in the title of Psalm 5 wasvirtually entirely predictable and, therefore, can scarcely count asdeliberate exposition. Again, the larger context must be attributed to thesource text, rather than to G.

It is true, of course, that the translator by doing what he did createdpotential for future exegesis; hence in reception history this potentialmight well be realized. Rösel, however, would have us believe thatalready at the constitutive stage much more was deliberately encoded inthe translated text. So he writes:

die Wiedergabe [verweist] nun auf ein weibliches Individuum. Damitlässt sich der Psalm als Lied einer Frau verstehen, die in ihrer Not zuGott ruft and auf seine Hilfe am Morgen hofft (V. 4).35.

Rösel is, of course, correct that the Greek text as it now stands has thepotential for such an interpretation, but that scarcely proves that theGreek translator himself had this in mind, and that, furthermore, hedeliberately reinterpreted the entire psalm in feminist terms. While it istrue that the psalm is a prayer, I see nothing in the Greek text that evenremotely makes reference to a female inheritor. Thus the observablefacts as well as the textual linguistic make-up of the translation as awhole testify to something far more mundane: G mistak-

35 M. Rösel, "Psalmenüberschriften" 131-32.

462

enly but by rather strict rules translated his source, and inadvertentlycreated a text radically at variance with the Hebrew at the phrase level.Finally, if the feminine inflection in Ps 5 indicates a woman, why not dothe same with th=j o0gdo/hj ("the eighth") in 6 and 11, seeing that it wasderived from the source text in the same manner?

Level 2.3:The use of u9pe\r tw~n lhnw~n for tytghA-l( in Psalms 8, 80(81),

83(84)). Here the observable facts are: (a) that again G fails tounderstand the Hebrew musical term; (b) that he derives it from tg("wine-press"); (c) that he construes the y as a w and thus ends up withthe plural; (d) that he isomorphically renders the entire phrase intoGreek. To be sure, interpretation perforce takes place, albeit based onignorance. But can we speak of exegesis or exposition? Moreover, canwe infer a cognitive process that denied that the psalms in question were"Gittith" but had to do with wine-presses instead? Rösel, however,wonders whether, since in prophetic literature lhno/j can connote adisplay of God's power, it might not connote the same in the psalms athand, since these are thought to be amenable to such an interpretation.Thus lhno/j, according to Rösel, should not be understood in its usualsense, even though that is what it normally carries both outside andinside the Septuagint, but should be understood metaphorically. Onceagain, that G had created a text with some potential, and that laterinterpreters might understand the Greek metaphorically, cannot bedenied. But as I see it, that is not relevant to the present discussion. Gproceeds literally, according to his analysis, and the larger context of thephrase is carried over from the source.

Level 2.4:The use of mh\ diafqei/rh|j for t#$xt-l) in 57(58), 58(59), 74(75).

Rösel makes no attempt at contextualizing this obscure phrase, adecision with which I fully agree. The phrase is nonetheless of interest,not for what G did with it but for what reception history was able to dowith it. Origen, for example, in comment on Ps 58:1, refers his readers toDavid's order to Abishai not to destroy Saul—the same phrase occurs inthe Greek—in 1 Rgns 26:9, when the two of them enter Saul's camp andcarry off the spear and water jug. And why not, since Psalm 58 is aDavid psalm and the superscription also refers to Saul's guarding David'shouse to kill him. But that is reception history not the constitutivecharacter of the translated text.

463

Or in Austermann's terms, Origen is writing an Auslegungstext notmaking an Übersetzung,36 and furthermore one of a formal-correspondence variety. What G does with the phrase in all three psalmsis predictable both on the verbal and nominal levels (verb 8x; noun 5x)and there is nothing in the psalms per se that lends support. But as acaveat against Origen's contextualizing of Ps 58, it should be noted thatneither of the other two have conducive detail in the superscriptions, and74 is not even a David psalm. Yet G derived the phrase in question fromhis source text in exactly the same manner as he did in 58.

Level 2.5:The use u9pe\r th=j a0ntilh/myewj th=j e9wqinh=j for rx#$h tly)-l(

in Psalm 21(22). Again I begin with the observable facts: (a) that thetranslator is familiar with Hebrew hly) ("doe") as is clear from18(17):34, 29(28):9; (b) that he did not know what to do with a doe inthe phrase at issue; (c) that he connects hly) with t w l y )("strength/help"); (d) that twly) occurs in 21:20 where he renders it bya0nti /lhmyij ; (e) that he then makes use of a0nti/lhmyij in thesuperscription; (f) that he introduces articles without formal warrant inthe Hebrew. There can be no doubt, therefore, that exposition at thephrasal level occurs in the process of translation. And given the fact thathe is unfamiliar with musical or liturgical terminology (including firstlines of songs) in the superscriptions, given his dislike for transcriptionsand, finally, given the fact that "concerning the doe of the morning"would make little if any sense even at the phrasal level, he did ratherwell. But the question that presents itself again is whether his concern formaking sense at the phrasal level means that he deliberately re-labels thepsalm as a whole. I can find no reason for such a conclusion. Even theChurch Fathers are surprisingly silent on this phrase. The only commenton it that I have been able to find is by Didymus the Blind who says thatit refers to a spiritual day that is being ushered in by "the sun ofrighteousness".37

Level 2.6:The use of ei0j a0na/mnhsin for rykzhl in Psalms 37(38), 69(70).

The observable facts of the case are: (a) that though the Hebrew √rkz ismost often in Psalms translated by the simplex Greek root mnh- (12x),

36 F. Austermann, Thesen zur Septuaginta-Exegese,” 383 (Thesis 6).37 Didymus, Comm. Psalmos, 23.

464

in the superscriptions to these two psalms as well as in 108(109):14 Guses the compound form; (b) that the exceptions cannot be explained bythe Hebrew stem (hiph‘il), since not all hiph‘ils are so translated; (c) thatno verbal noun of the simplex form is attested. In light of (c) it may wellbe that G's option is determined linguistically rather than semantically.Rösel (133) sees significance in two things: (a) that the phrase can beused in a cultic context and b) that the Hebrew infinitive is translated byei0j + a verbal noun, rather than by an infinitive. Even if we grant Röselthe cultic use of the Greek phrase, we would still have to conclude thatno deliberate interpretation took place in the translation process, sincethe source text would already have had that sense. Rösel's second point,it seems to me, is purely linguistic. That is to say, according to G'sstandard practice for infinitives with preposed l, rykzhl would haveproduced tou= a0namnhsqh=nai. But had he followed standard practicehere, "Dauid" would have had to function as its subject. Hence theresultant text would be "A Psalm. Pertaining to Dauid in order that hemight commemorate." If one then further regards peri\ sabba/tou inPsalm 37 as original text (which I do not), one would end up withDavid's being told that he should remember about the sabbath. A similarproblem would arise in Psalm 69 if our translator had rigidly stuck to hisstandard equivalent, and in doing so had perforce created a subject of theinfinitive.38 Thus if the translator was intent on safeguarding what theHebrew text is thought to mean, to use the purpose infinitive was not arealistic option for him. Thus G is not going beyond the Hebrew at all,except for the fact that rykzh may mean "memorial offering" (soN R S V ) , while a 0 n a / m n h s i j s i m p l y m e a n s"remembering/recalling/commemorating." Thus even if one were toapply the Greek phrase to the psalm as a whole no exposition or exegesiswould have taken place beyond what the source text already gives us.

Linguistically precisely the same phenomenon occurs in Psalm59(60) where ei0j didaxh/n (ei0j + verbal noun) is used to translate dmll(an infinitive), and all three (Psalms 37, 59, 69) may be contrasted withtou= a0pokriqh=nai in 87, discussed on p. 457.

38M. Rösel, "Psalmenüberschriften" 133. The two references Rösel cites in support (3Rgns 17.18 and Am 6.10) demonstrate my point.

465

Level 2.7:The use of u9pe\r tou= a0gaphtou= for tdydy in Psalm 44(45)). The

facts of the case are: (a) that G once again fails to understand the importof the Hebrew (Cf. NRSV "A love song"); (b) that he (correctly as itseems) derived the word from the Hebrew adjective dydy ("beloved");(c) that G ignores the final t; d) that he renders dydy in the same way herenders it all four other times in Psalms (60[59]:7, 64[83]:2, 108[107]:7,127[126:2) by a0gaphto/j Rösel39 is right in noting that both u9pe/r andthe article are unwarranted by the Hebrew. Hence some interpretationtakes place at the phrasal level. Whereas the Hebrew according to G'sanalysis would mean "a song of a beloved" the Greek would mean "onbehalf of the beloved" (u9pe/r is common in dedicatory statements). Butthen Rösel links "beloved" here with "beloved" in 67(68):13 where it infact is used twice for Hebrew Nwddy. So what happens is that the latter is

equated with dydy, since he evidently doesn't know what to do withNwddy. Given the fact that dydy is consistently glossed with a0gaphto/j,the linguistic connection G forges in 68(67):13 based on ignorance of hissource text is understandable, and expositional on the word or phraselevel. Furthermore, that same linguistic connection is made with all otheroccurrences of dydy. But can one then also argue, as Rösel does, thatsince in his judgment tou= a0gaphtou (=Nwdd) in 67(68):13 refers toGod,40 it does as well in 44 on the grounds that there too the singularoccurs? But since singular and plural occur (for this word) in lockstepwith the source text, how can this be deemed expositional in anymeaningful way?41 There can be little doubt that such intra-textualexegetical connections would be made in reception history, but is italready encoded at the constitutive stage? When one approaches the textfrom within and bases oneself on its linguistic make-up, all suchinstances appear as purely linguistically based, and expositional purelyon the phrasal level. G etymologizes what he doesn't understand andrefuses to transcribe and in so doing creates a text that differs moreradically from the Hebrew than would have been the case if he hadunderstood the Hebrew. The rest was up to reception history.

39 “Psalmenüberschriften,” 133.40 This interpretation itself is based on creative contextualizing.41 Rösel further states that the reference in 67:13 is "eindeutig eschatologisch," 134 note53.

466

Level 2.8:The use of u9pe\r tw~n a0lloiwqhsome/nwn for Myn#$#$-l( in Psalms

44(45), 59(60), 68(69), 79(80). The observable facts of the case are: (a)that G did not understand his source text and derives the Hebrew from√hn#$ ("change"); (b) that he analyzes the form as a non-feminine pluralparticiple of that verb; (c) that the left over initial #$, like the m inpreceding jxcnml, he represents by the Greek article; (d) that in so doinghe maintains an isomorphic relationship to the source text; (e) that in34(33):1 and 77(76):11 G translates √hn#$ with a0lloi- ("change"),which is in fact the standard equation in the Septuagint; (f) that since√hn#$ occurs most often in Daniel, a0lloio/w most often occurs there; (g)that most often throughout the Septuagint a0lloio/w has a non-eschatological sense. Rösel, however, goes two steps beyond this. First,to him, the phrase, together with preceding ei0j to\ te/loj, is "gewisseschatologisch."42 Second, the phrase makes the psalms in question intoeschatological psalms. That at the phrasal level G engages in expositionis clear from the fact that, although the Greek-Hebrew equation as suchis predictable, the use of the future passive participle is not. Thus time-subsequent and passive transformation is being signaled ("those that willbe changed"). But even if one were to grant that the word here has asense it normally does not have, there is no other indication that it wasmeant to function beyond the phrasal level, except that one of the fourpsalms in question (44[45]) can be interpreted eschatologically. But whatabout the other three? The Greek-Hebrew equation, even though itcannot be predicted on the basis of Psalms alone, nevertheless turns outto be predictable in light of the Septuagint as a whole.

That the phrase in question would be read eschatologically beyond itsown boundaries in Christian reception history, the more since, in itssuperscription, Psalm 44 also features u9pe\r tou= a0gaphtou=, wasinevitable. It comes, therefore, as no surprise that Athanasius, forexample, says that "the beloved" is David's son, Christ, and the phrase inquestion refers to the a0lloi/wsij brought about by the advent of Christ.Cyril, on the other hand, has our phrase refer to Jews and Greeks who,according to Paul, in Christ became one beloved people (cf. Rom 9:25).

42 “Psalmenüberschriften,” 134. Didymus the Blind, for example, does the same thing(Commentarii in Psalmos 40-44, 336) and, furthermore, brings in Ps 76:11.

467

I close the present discussion with two phrases which Rösel accordsspecial treatment because of their allegedly even clearer eschatologicalimport, namely, sune/sewj / ei0j su/nesin and ei0jto\ te/loj.

Level 2.9:The observable facts on sune/sewj / ei0j su/nesin for lyk#&m are as

follows: (a) that G did not understand lyk#&m as a certain type of song;(b) that he derived the term from the verb √lk#& ("be prudent"); (c) thatin Psalms he translated √lk#& with suni/hmi + cognates some 22 times;(d) that suni/hmi + cognates is used to translate √Nyb ("understand") some27 times; (e) that in most superscriptions he translates lyk#&m with agenitive (32[31], 52[51], 53[52], 54[53], 55[54], 74[73], 78[77], 88[87],89[88], 142[141]); f) that in three superscriptions he renders it by ei0jsu/nesin (42[41], 44[43], 45[44]).

To be sure this summary points up some interesting facts. Forexample, G vividly demonstrates his lack of familiarity with lyk#&m as atype of song in 47(46):8 where he translates lyk#&m wrmz ("play aMaskil") as ya/late sunetw~j ("make music [on strings] withunderstanding"). Similarly, it is interesting that in all cases he pushessu/nesij from the word level to the phrase level, either by inflection orby preposing a preposition, though perhaps it deserves noting that verbalnouns in the superscriptions are regularly made to function at the phrasallevel whether or not there is explicit warrant in the Hebrew, the only twoexceptions being sthlografi/a in Ps 15 and ai1nesij in Ps 144. Thusthe reason for turning su/nesij into a phrase may be chiefly linguistic.Whatever the precise reason, exposition at the phrasal level has occurred.Beyond that, if perchance G opted for ei0j su/nesin (as a purposeexpression) because of the adjacent phrase toi=j ui9oi=j Kore, on theassumption that G thought that the latter could do with a bit ofunderstanding (cf. Num 16), we can even say that an expositional moveextended to the propositional level. Rösel (136-37), however, wants topush it well beyond that point, since for him it re-labels the entire psalmwhenever su/nesij occurs in the superscription as a gloss for lyk#&m.That seems highly questionable since its occurrence is once againpredictable on the basis of the source text, and similarly, on the fewoccasions that a member of the suni/hmi group occurs within the psalmitself (31:8. 9, 52:3, 77:72), it is again predictable on the basis G'sstandard equations.

468

That being the case, how can it be argued that the translator is engaged indeliberate interpretation, i.e. exposition? All that can be said is that, sinceHebrew √lk#& and Greek su/nesij + cognates do not have an identicalsemantic range, interpretation may be taking place in the translationalprocess. Rösel, however, takes yet another step, since he writes:

Das fragliche Nomen ist nun mitsamt dem zugehörigen Verbumsuni/hmi in der Jesaja-LXX wie in der Dan-LXX eindeutig im Sinneeines eschatologisch-apokalyptischen Verstehens der Wege Gotteskonnotiert; man erinnere sich nur an die berühmte Übersetzung vonJes 7,9 mit "glaubt ihr nicht, so versteht ihr nicht".43

He then proceeds to certain passages in the Psalter where su/nesij or acognate thereof might carry the same sense, for example: Psalms15(16):7; 48(49):13, 21;146(147):5 and 110(111):10. Thus Rösel'sargument here is effectively that, since su/nesij elsewhere in the LXXcan have an eschatological-apocalyptic sense, it should be given thatmeaning whenever a given text can bear it. But that ignores twofundamental facts: that suni/hmi + cognates, both without and within theLXX, rarely carries that meaning, and furthermore, that in three of thefour passages he cites in the Psalter the Greek word is predictable. Thatleaves 15(16):7 where suneti/zw ("to make to understand") translatesHebrew C(y ("to give counsel"). Since in this case suneti/zw is a non-default rendering for C(y (= bouleu/omai 4x, e0pisthri/zw 1x, suneti/zw1x), it of course attracts exegetical interest; but it scarcely gives it aneschatological-apocalyptic meaning. That su/nesij anywhere has such ameaning is quod est demonstrandum. Similarly, that what G does hasmore than an indirect and non-deliberate effect on the psalms in questionis equally quod est demonstrandum.

Level 2.10:Perhaps the most lavish interpretation Rösel reserves for ei0jto\

te/loj, a phrase that occurs in the superscriptions more often than anyother (ca. 55), with the exception of tw~| Dauid (ca. 73). The observablefacts are: (a) that ei0jto\ te/loj and xcnml form a closed Greek-Hebrewequation; (b) that G was unfamiliar with the meaning "leader" (NRSV) or"director" (BDB); (c) that G arrived at his translation via his equation ofei0jte/loj with xcnl. As in the case of su/nesij, Rösel would have usbelieve that ei0jto\ te/loj should be understood

43"Psalmenüberschriften," 136.

469

eschatologically. He briefly entertains others possibilities but thenwrites:

Sinnvoller ist die Übersetzung mit "Ende", die man wohl auf dieEndzeit beziehen muss; die entsprechenden Lieder zielen demnachauf die Endzeit. Diese Überlegung wird durch die auffälligeVerwendung des Artikels unterstützt, die m. E. eindeutig auf einbestimmtes Ende zielt.44

If such a claim could be substantiated it would mean that our Greektranslator in the act of translating has made some 55 psalms into psalmsabout the end time. But the argument that leads to such a conclusionseems fatally flawed. I begin with Hebrew xcn for which te/lojregularly serves as a gloss. According to the lexica it would seem safe tosay that the root has essentially three components of meaning: "(pre-)eminence, successfulness, perpetuity." It is thus little wonder that xcnlis commonly glossed in English as "forever," that is to say, "inperpetuity." Though te/loj can have a great many meanings and clearlyhas considerable semantic overlap with xcn, the component not coveredvery well, if at all, by te/loj is that of perpetuity, i.e. the temporaldimension. This becomes at once clear when one investigates how xcnis translated in the Septuagint. Outside of the Psalter the root occurssome 35 times: five times one finds ei0j te/loj ("completely" Hab 1:4;Job 4:20, 14:20, 20:7, 23:7), five time ei0j ni=koj ("victoriously" 2 Sam2:26; Jer 3:5; Amos 1:11, 8:7; Lam 5:20) + tou= nikh=sai ("to winvictory" Hab 3:19), and h9 ni9kh ("victory" 1 Chron 29:11).45 Seeminglyrelated to the concept of "victory" are i0sxu/w ("to be powerful/prevail")in Isa 25:8, katisxu/w ("to prevail over") in Jer 15:18, and e0nisxu/w ("toprevail in") in 1 Chron 15:21. And again trading on the notion of pre-eminence are glosses like e0rgodiw&kthj ("taskmaster") in 1 Chron 23:4and 2 Chron 2:17, as well as e0piskope/w ("to oversee") in 2 Chron34:12. Thus there is plenty that reflects the components of "(pre-)eminence" and "successfulness." Interestingly, however, when thecomponent of "perpetuity" comes into play xcn is glossed by temporalphrases: ei0j to\n ai0w~na (Isa 28:28; Jer 27[50]:39), ei0j to\n ai0w~naxro/non (Isa 13:20, 33:20), xro/non polu/n (Isa 34:10), dia\ panto/j (Isa57:16) and e1ti (Job 34:36).

Thus one can conclude with reasonable assurance that outside of thePsalter te/loj does not seem to have a temporal dimension. Yet that is

44 “Psalmenüberschriften,” 13845 Aquila and Quinta use ei0j ni=koj for xcnl.

470

precisely what Rösel claims for the Psalter in his lead-up to "dieEndzeit."46 To prove his point he makes reference to three passages in thePsalms where ei0j te/loj appears as a parallel to ei0j to\n ai0w~na (9:19,76:8f[?], 102:9). The inference is, therefore, that "parallel" means"identical." That seems to me problematic. One can in fact argue that inPsalms too te/loj is not perceived to have a strictly temporal dimension,since in Ps 49(48):20 where the Hebrew has xcn-d( and where themeaning is patently temporal, G switches to e#wj ai0w~noj. Since this is anon-default rendering of xcn it can be taken to have some exegeticalsignificance.

Rösel's proposal to read ei0j to\ te/loj eschatologically raises a bynow familiar problem. In non-philosophical Classical and Hellenisticliterature te/loj as a nominal means nothing more often than"conclusion" (natural or logical) and as an adverbial it means nothingmore frequently than "in conclusion" or "completely/finally," with nomore of an eschatological overtone than the English glosses I have used.Polybius, for example, regularly uses ei0j te/loj, Similarly, within theSeptuagint (some 94 occurrences according to Hatch-Redpath, notcounting the Psalter) te/loj rarely has an eschatological sense. In light ofall that, with what justification can the claim be made that the phrase ei0jto\ te/loj has an eschatological sense and is thus an exegeticalcontribution of the translator. Is it because of the article, which Röselsees as supporting such a claim? But the article is there simply tomaintain isomorphism with the source text, and perhaps moreimportantly to allow G to reproduce a contrast in his source text: ei0jte/loj = xcnl and ei0j to\ te/loj = xcnml while deriving both from thesame root.

That the Fathers of the Church, who read the entire Septuagint as apraeparatio euangelica would read ei0j to\ te/loj and in fact te/lojgenerally from an eschatological perspective is of course true. So, forexample, Asterius the Sophist in comment on Ps 9:1 exclaims:

What is to; te/loj? The beginning of the proclamation of theGospel, which is the te/loj of the Law and the Prophets (ti/ to\te/loj; h9 a0rxh\ tou= eu0aggelikou= khru/gmatoj, o3 e0sti te/loj tou=no/mou kai\ tw~n profhtw~n).

In similar vein 1 Pet 4:7 writes that "the end of all things is near"(pa/ntwn de\ to\ te/loj h1ggiken). But to superimpose such a meaningonto the Septuagint runs afoul of what I consider to be a basic and vi-

46 “Psalmenüberschriften,” 138.

471

tal distinction between the chronologically oldest and logically priorSeptuagint, on the one hand, and its reception history, on the other. Whatthe original text meant has to be determined on the basis of itsconstitutive character.

Level 3: Interpretation at the Sentence LevelIn earlier comment on ei0j su/nesin in Psalms 41, 43, and 44 I have

already suggested that the structure of the phrase may have beendetermined by the preceding phrase toi=j ui9oi=j Kore. If that is the casewe can speak of contextualization from the phrasal to the propositionaland, therefore, to the clausal or sentence level.

As an example of intra-clausal exegetical activity one might citePsalm 3: o9po/te a0pedi/drasken a0po\ prosw&pou Abessalwm tou= ui9ou=au0tou= for wnb Mwl#b) ynpm wxrbb. Though the grammaticalinformation is transferred almost isomorphically to the Greek, it is ofinterest that in the case of the Hebrew infinitival construction G opts foro9po/te plus an imperfect verb. Since both the conjunction and theimperfect indicative verb are uncommon and therefore marked items inthe Psalter as well as the Septuagint corpus, one can infer a certaindeliberateness on the part of the translator. So here he portrays David'sflight from his son as a withdrawal in progress, something the Hebrewdoes not show explicitly. But since the information conveyed by thisclause was already in the parent text, G cannot be said to havecontextualized the sentence at the paragraph level, i.e. the entire psalm oreven a part thereof. What exposition he did, he did purely within thesentence.

As I have suggested elsewhere,47 there can be no doubt that all such"historical" superscriptions played an important exegetical role in thetransmission history of the Book of Psalms, both before and after theywere translated into Greek. But if our interest lies in the specificcontributions of the Greek translator to this history of exegesis, it mustbe ascertained whether or not such items—be they word, phrase, orsentence—were introduced as part of the translational process. To theextent that such items are also attested by the Masoretic Text, one cansafely assume that they were inherited by G from his source text. Thatbeing the case, he can be given expositional credit only for whatexposition he can be shown to have accomplished at the sentence level.

For a final possible example one might turn to Psalm 55(56) u9pe\rtou= laou=

47 A. Pietersma, "Exegesis and Liturgy," 99-138.

472

tou= a0po\ tw~n a9gi/wn memakrumme/nou for Myqxr Ml) tnwy-l(. Atthe phrasal level G does not know what to do with a "dove" (hnwy) anymore than he knows what to do with a "doe" in Psalm 21(22). As aresult, here as in 21(22) he comes up with something that makes sense atleast within the phrase.48 Though it is likely that G did not stray very farfrom the consonantal text, one wonders whether the sense he gave thephrase is related to the last clause in the superscription which states thatDavid was in a foreign land, away from Israel's shrine. If that is so wehave here another instance of clausal and phrasal contextualization.Since as in the previous example the items as such were already in theparent text, G cannot be credited with exposition beyond the sentencelevel.

Level 4: Interpretation at the Paragraph LevelAt this level of interpretation significant exposition of the source text

clearly takes place, and like all other levels of interpretation it too can befound in the translated corpus. In connection with the superscriptions,one naturally thinks of the superscriptions in the Greek which are lackingin the Masoretic text. But as I have already noted, if our interest lies inthe contribution of the translator, that is, in the constitutive character ofthe translated text, not only do we have to remove from considerationitems G inherited from his source text, but also items that belong to thereception history of the Greek text. In an earlier article I have dealtextensively with this issue. Here a single example must suffice.

While in MT the superscription of Psalm 27(26) is a simple dwdl,the Greek text adds pro\ tou= xrisqh=nai ("before he [David] wasanointed"). What happened here in the reception history of the Book ofPsalms is reasonably clear. From being simply a "David psalm," itbecame a psalm associated with a particular period in David's life,namely, before he was anointed king over Judah (2 Sam 2:4) and overIsrael (2 Sam 5:3). The impetus for the addition arose from the Greektext of v. 5:

o3ti e1kruye/n me e0n skhnh=| e0n h9me/ra| kakw~n mou:e0ske/pase/n me e0n a0pokru/fw| th=j skhnh=j au0tou=,e0n pe/tra| u3ywse/n me:

48 For a discussion of G's possible misreadings of the consonantal text see MartinFlashar, "Exegetische Studien zum Septuagintapsalter," ZAW 32(1912) 244.

473

For he hid me in the tabernacle in the day of my troubles;he sheltered me in a secret spot of his tabernacle;he set me high on a rock. (NETS)

This verse was thought to refer to David's stop-over at the tabernacle atNob (1 Sam 21), an event which unmistakably predated his becomingking. Though theoretically the extra clause could have been part of G'ssource text, this becomes unlikely once one realizes that the terms for"tabernacle" in this verse are hks and lh) , but not Nk#$m. Whilelinkage with the tabernacle would not be impossible within Hebrewtransmission history, the Greek text makes it all but inevitable, since inGreek skhnh/ is the standard term for the old desert shrine. The crucialquestion becomes, however, whether it was the translator who added theexegetical note based on information supplied in v. 5, or whether it wasthe reception history of the Greek text that was responsible. Even thoughthere is evidence to suggest that G reserved skhnh/ for the tabernacle(hence the NETS translation), I consider it more likely that the piece ofexegesis belongs to the history of interpretation of the Greek Psalter(hence the square brackets in NETS). Be that as it may, for my presentpurpose suffice it to say that if the clause is attributable to G, we haveclear evidence that G at times engaged in exposition at the paragraphlevel, that is, the psalm as a whole. If, on the other hand, the clause stemsfrom reception history, it ought not be cited as evidence for exegesis inthe Septuagint itself.49

CONCLUSION

I have sought to argue that though genuine exegesis and expositioncan be found in the Septuagint, including in the Greek Psalter, it needs tobe identified on the basis of its textual-linguistic make-up. If its textual-linguistic make-up argues for a translation characterized more by formalcorrespondence than by dynamic equivalency, one's approach tohermeneutics in the Septuagint should be governed by these findings. AsI see it, that means at a minimum that exegesis needs to demonstrated,not presupposed. From that perceptive I would suggest that one workfrom the least intelligible phenomena to the more intelligible; that oneproceed from the word level to higher levels of constituent structure; thatone pay more attention to the translator's deviations from his Hebrew-Greek defaults than to his defaults and stan-

49 For another instance of exegesis beyond the sentence level, see Boyd-Taylor “APlace in the Sun,” 77-105.

474

dard equations or, to put it differently, that greater weight be given towhat is unpredictable than to what is predictable; that one assign greatercontext to segments of the Greek text than to the corresponding segmentsof the Hebrew text only as a last resort.

To read the translated text in the light of its constitutive character isone thing, but to read it in the light of a culturally reassigned functionand position is quite another.

SELECT BIBLIOGRAPHY

Austermann, F. "Von der Tora im hebräischen Psalm 119 zum Nomosim griechischen Psalm 118," in E. Zenger (ed.), Der Septuaginta-Psalter (Freiburg: Herder, 2001) 331-47.

—. "Thesen zur Septuaginta-Exegese am Beispiel der Untersuchung desSeptuaginta-Psalters," in A. Aejmelaeus and U. Quast (eds.), DerSeptuaginta-Psalter und seine Tochterübersetzungen. (Göttingen:Vandenhoeck & Ruprecht, 2000) 380-86.

Barr, J. The Semantics of Biblical Language. Oxford 1961 (SCM edition1983).

—."Common Sense and Biblical Language," Biblica 49 (1968)337-87.

Boyd-Taylor, C. "A Place in the Sun: The Interpretative Significance ofLXX-Psalm 18:5c," BIOSCS 31(1998) 71-105.

Brock, S. “The Phenomenon of the Septuagint,” OTS 17 (1972) 17.Brown, W. P."Reassessing the Text-Critical Value of the Septuagint-

Genesis 1: A Response to Martin Rösel," BIOSCS 32 (1999) 35-39.Flashar, M. "Exegetische Studien zum Septuagintapsalter," ZAW 32

(1912) 81-116, 161-89, 241-68.Hendel, R. S. "On the Text-Critical Value of Septuagint Genesis: A

Reply to Rösel," BIOSCS 32 (1999) 31-34.Joosten, J. "Exegesis in the Septuagint Version Hosea," in J. C. De Moor

(ed.), Intertextuality in Ugarit and Israel (OTS 40.) (Leiden: Brill,1998) 62-85.

Kooij, A. van der. "Zur Frage der Exegese im LXX-Psalter. Ein Beitragzur Verhältnissbestimmung zwischen Original und Übersetzung," inA. Aejmelaeus and U. Quast (eds.), Der Septuaginta-Psalter undseine Tochterübersetzungen. (Göttingen: Vandenhoeck & Ruprecht,2000) 366-79.

Pietersma, A. Review of Schaper: Bibliotheca Orientalis 54 (1997)185-90.

—.The Psalms, in A. Pietersma and B. Wright (eds.), A New EnglishTranslation of the Septuagint. New York and Oxford: OxfordUniversity Press, 2000.

—."Exegesis and Liturgy in the Superscriptions of the Greek Psalter," in

475

B. A. Taylor (ed.). X Congress of the International Organization forSeptuagint and Cognate Studies, Oslo, 1998. (SBLSCS 51; Atlanta:Society of Biblical Literature, 2001) 99-138.

—."The Present State of the Critical Text of the Greek Psalter," in A.Aejmelaeus and U. Quast (eds.), Der Septuaginta-Psalter und seineTochterübersetzungen. (Göttingen: Vandenhoeck & Ruprecht, 2000)12-32.

Rösel, M. "Die Psalmüberschriften des Septuaginta-Psalter," in E.Zenger (ed.), Der Septuaginta-Psalter (Freiburg: Herder, 2001) 125-48.

—. "The Text-critical Value of Septuagint-Genesis," BIOSCS 31 (1998)62-70.

Schaper, J. Eschatology in the Greek Psalter. (WissenschaftlicheUntersuchungen zum Neuen Testament. 2. Reihe 76; Tübingen:Mohr Siebeck, 1995).

Thesaurus Linguae Graecae (TLG). CD ROM# e. 1999.Toury, G. Descriptive Translation Studies and beyond. (Amsterdam and

Philadelphia: Benjamins, 1995).